注水站故障诊断智慧模型研究  

Research on Intelligent Model for Fault Diagnosis of Water Injection Station

作　　者：徐加军[1] 张元法 李永强[1] 李荣权[1] 

机构地区：[1]中石化胜利油田分公司胜利采油厂,山东 东营

出　　处：《石油天然气学报》2022年第2期81-91,共11页Journal of Oil and Gas Technology

摘　　要：针对注水泵站目前系统存在数据不可靠、不能完全替代人工巡检、缺少智能管控功能等现状,有必要开展故障诊断智慧模型研究,通过对胜利采油厂离心泵注水站设备运行模型研究与构建,准确识别注水站设备脆弱部分,对其进行重点监测和控制。在已建立的注水站信息化数据基础上,分析数据完备性、准确性和可替代性,首先对关键数据进行清洗,排除异常数据,对传感器数据进行准确性判别,再对异常数据进行筛查,结合卷积神经网建立设备故障智能判别模型。根据建模利用软件开发技术实现一套注水站故障诊断智慧平台,提高注水站故障诊断率,达到提升系统可靠性和智能化水平的目的。It is necessary to carry out the research on the intelligent model of fault diagnosis because we cannot completely replace manual patrol inspection and lack of intelligent control function in the system of water injection pump station in view of the current situation of unreliable data. Through the research and construction of the equipment operation model of centrifugal pump water injection station in Shengli oil production plant, accurately identify the vulnerable parts of water injection station equipment and focus on monitoring and control. Based on the established information data of water injection station, analyze the completeness, accuracy and substitutability of data. Firstly, clean the key data, eliminate the abnormal data, judge the accuracy of sensor data, screen the abnormal data, and establish the intelligent judgment model of equipment fault combined with convolution neural network. According to the modeling, a set of intelligent platform for fault diagnosis of water injection station is realized by using software development technology to improve the fault diagnosis rate of water injection station and achieve the purpose of improving the reliability and intelligence level of the system.

关 键 词：注水站 卷积神经网络 故障诊断 

分 类 号：TE357.6[石油与天然气工程—油气田开发工程]